These potentials are much higher than those found in earlier experiments. The method which was usually adopted to detect impurities, with a spectroscopic tube, is not sufficiently sensitive to indicate small traces of impurities which may have a considerable effect on the sparking potentials.

LXXXVII. The Crystal Structure of Cu,Al. (8 CopperAluminium.) By Dr. A. J. BRADLEY.

THE

[Plate XV.]

THE alloy system Cu-Al has been investigated thermally and microscopically by several independent investigators. The most recent work is that of Stockdale †, who has also made a special study of the alloys rich in copper ‡. X-ray methods have been employed by several workers §, the most complete investigation by this method being that of Jette, Westgren, and Phragmèn . The results obtained by the latter do not entirely agree with the equilibrium diagram proposed by Stockdale for alloys with between 16 and 30 per cent. of aluminium. We have recently made an examination of some alloys in this range, which confirms neither of the above results, but corroborates the results of further work, done recently by Westgren and Phragmèn, and communicated privately to us.

The most important point settled by this recent investigation is that Stockdale's 8-phase actually only extends over the range between 16 and 19 per cent. of aluminium. Beyond 19 per cent. of aluminium the photograms become rather difficult to interpret, and it was evident that the most satisfactory way to tackle the problem was to start by determining the way the atoms were arranged in the 8-phase, with between 16 per cent. and 19 per cent. of aluminium.

The clue to the crystal structure of the 8-phase is given by the close resemblance between the photograms of this phase and those obtained from alloys of copper and zinc containing between 61 and 69 per cent, of zine. The structure of the

*Communicated by Prof. W. L. Bragg. F.R.S.

† Journ. Inst. Metals, xxxi. p. 275 (1924).

Journ. Inst. Metals, xxviii. p. 273 (1922).

§ Owen & Preston, Proc. Phys. Soc. Lond. xxxvi. p. 14 (1923); Jette, Westgren, & Phragmèn, Journ. Inst. Metals, xxxi. p. 193 (1924). Loc. cit.

latter alloys was found by Bradley and Thewlis*, and in the present paper an attempt has been made to trace out the relationship between the two structures.

Data.

Jette, Westgren, and Phragmèn (loc. cit.) obtained single crystals of alloys with 16 per cent. of aluminium, large enough to use for a Laue photograph. This showed that the crystal symmetry was cubic, the class being T, O, or O1.

We have taken powder photograms of various alloys within the range 16-19 per cent. aluminium, and find that all the lines fit in with the assumption that the structure is cubic. The results obtained from two photograms are given below, in Table I. Copies of these photograms are shown in Pl. XV. together with a film of copper-zinc taken

for comparison. In each case the radiation employed was obtained from a hot cathode tube with a copper anticathode, the B-radiation being cut off by the use of a nickel screen. Only one B-line can be observed on each film, corresponding to the strongest a-line. At the ends of the photograms the a-doublet is clearly resolved. Only measurements from lines which are resolved have been used for the determination of the lattice dimensions. By this means it has been possible to get a high degree of accuracy in these calculations.

The number of atoms per unit cell was found to be the same in the copper-aluminium structure as in the copper-zine structure, namely, 52. The data from which this figure was obtained are given in Table II.

values from the lines where the a-doublet is resolved.

Derivation of the Structure of & Cu-Al from the
7 Cu-Zn Type.

The similarity between the copper-aluminium and copperzine photograms, which has been pointed out by Westgren and Phragmèn, is so great that there can be no doubt that the structures are very closely related. The differences are, however, sufficiently marked to show that the structures are not quite the same in every respect. By carefully studying the difference it has been found possible to assign a structure to & Cu-Al by a slight modification of the copper-zinc structure. The structure so obtained has the requisite symmetry shown by the Laue photograph. Starting out from the assumption that the structure of 8 Cu-Al is based on the y Cu-Zn type of structure, a variety of possible modifications suggest themselves. A choice between these possibilities has been made, taking the intensities of the lines as the sole guide for this purpose. The validity of the structure ultimately found rests on the excellent way in which it explains the observed intensities, including the differences in intensity of corresponding lines on the copperaluminium and the copper-zine photograms.

The 52 atoms in the unit cube of the copper-zinc type of structure are divided among four groups of structurally equivalent atoms, which may be called A, B, C, and D atoms

respectively. There are eight A atoms, eight B atoms, twelve Catoms, and twenty-four D atoms. The A and D atoms are zinc, the B and C atoms are copper, so that that there are 32 zinc atoms and 20 copper atoms. The relative proportions of the two constituents in the copper-aluminium alloy is quite different, there being about twice as many copper atoms as aluminium atoms. The distribution of the atoms in & Cu-Al must therefore be quite different from that in the y Cu-Zn γ alloy.

This fact gives a possible explanation of the differences in the intensities in the photograms of the two alloys. The photograms differ in two ways. The relative intensities of the lines on the y Cu-Zn photogram are not in every case the same as those of corresponding lines on the & Cu-Al photograms. There are, moreover, some additional lines on the Cu-Al photograms which, for the most part, belong to planes with h+k+l odd, indicating that the unit cube is primitive and not body-centred, as in the case of y-brass.

These intensity changes cannot be accounted for by supposing that the positions of the atoms have suffered a slight displacement, the effect of which would be to produce a slight change in the least deviated reflexions and a much larger change in the most deviated reflexions. This is just the reverse of what one observes. It is therefore safe to conclude that the atoms, as a whole, occupy almost identical positions in the two alloys. The intensity changes must be ascribed to the differences in the distribution of atoms of the two species, and they can therefore be utilized in order to find how the copper and aluminium atoms are distributed.

The first condition imposed by the reflexions is that the unit cube of Cu-Al must be primitive. In order to see how this change from the y Cu-Zn type can be brought about, it is only necessary to consider what arrangement is responsible for the y Cu-Zn structure being body-centred. The y Cu-Zn structure is body-centred because there is for every atom with coordinates a yz, a corresponding atom with coordinates+y+z+, and each of these pairs of atoms is of the same kind. If they were of different kinds the unit would be primitive. Thus the unit cube of the Cu-Al structure will be primitive, provided that, in at least one of the four groups of atoms, an atom with coordinates a y z is aluminium and an atom with coordinates + \ y + ↓ ÷ + ↓ is copper. In other words, each group of atoms, A, B, C, D, may itself consist of two structurally independent sets of atoms. There are, therefore, eight possible independent sets of atoms in the 8 Cu-Al structure.